Mechanism and properties of positive allosteric modulation of N-Methyl-D-

Aspartate receptors by 6-alkyl 2-naphthoic acid derivatives

Kiran Sapkotaa, Mark W. Irvineb, Guangyu Fangb, Erica S. Burnellb, Neil Bannisterb,1,

Arturas Volianskisb,c, Georgia R. Culleya2, Shashank M. Dravidd, Graham L.

Collingridgeb,3, David E. Janeb*, Daniel T. Monaghana*

a Department of Pharmacology and Experimental Neuroscience, University of Nebraska

Medical Center, Omaha, Nebraska 68198-5800 U.S.A. bCentre for Synaptic Plasticity, School of Physiology, Pharmacology and Neuroscience,

Biomedical Sciences Building, University Walk, University of Bristol, Bristol,

BS8 1TD, UK cCentre for Neuroscience and Trauma, Blizard Institute, Barts and The London School of

Medicine and Dentistry, Queen Mary University of London, UK dDepartment of Pharmacology, Creighton University, Omaha, Nebraska

1University of Bath, Claverton Down, Bath BA2 7AY

2Department of Physiology, Institute of Neuroscience and Physiology, The Sahlgrenska

Academy, Gothenburg University, Medicinaregatan 11, 405 30, Gothenburg,

Sweden.

3Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital, Toronto, Ontario M5G

1X5, Canada., Department of Physiology, University of Toronto, Toronto,

Ontario, M5S 1A8.

* DEJ and DTM contributed equally as senior authors.

Corresponding author: DTM, [email protected] Keywords

L-glutamate

N-methyl-D-aspartate

Potentiator

Positive allosteric modulator

Deactivation

Ligand-binding domain

1 Abstract

The theory that N-methyl-D-aspartate (NMDAR) hypofunction is responsible for the symptoms of schizophrenia is well supported by many pharmacological and genetic studies. Accordingly, positive allosteric modulators (PAMs) that augment NMDAR signaling may be useful for treating schizophrenia. Previously we have identified several

NMDAR PAMs containing a carboxylic acid attached to naphthalene, phenanthrene, or coumarin ring systems. In this study, we describe several functional and mechanistic properties of UBP684, a 2-naphthoic acid derivative, which robustly potentiates agonist responses at each of the four GluN1a/GluN2 receptors and at neuronal NMDARs.

UBP684 increases the maximal L-glutamate/ response while having minor subunit-specific effects on agonist potency. PAM binding is independent of agonist binding, and PAM activity is independent of membrane voltage, redox state, and the

GluN1 exon 5 N-terminal insert. UBP684 activity is, however, markedly pH-dependent, with greater potentiation occurring at lower pHs and inhibitory activity at pH 8.4.

UBP684 increases channel open probability (Po) and slows receptor deactivation time upon removal of L-glutamate, but not glycine. The structurally related PAM, UBP753, reproduced most of these findings, but did not prolong agonist removal deactivation time.

Using cysteine mutants to lock the GluN1 and GluN2 ligand-binding domains (LBDs) in the agonist-bound states, we found that PAM potentiation requires GluN2 LBD conformational flexibility. Together, these findings suggest that UBP684 and UBP753 stabilize the GluN2 LBD in an active conformation and thereby increase Po. Thus,

UBP684 and UBP753 may serve as lead compounds for developing agents to enhance

NMDAR activity in disorders associated with NMDAR hypofunction.

2 I. Introduction

The primary excitatory neurotransmitter in the vertebrate CNS, L-glutamate, activates three distinct families of ligand-gated ion channel receptors that are named for agonists by which they are selectively activated, N-methyl-D-aspartate (NMDA), (S)-2- amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) and kainate (Monaghan et al., 1989; Watkins and Evans, 1981; Watkins et al., 1990). While AMPA and kainate receptors underlie fast excitatory synaptic transmission in the CNS, NMDA receptors

(NMDARs) activate relatively slow currents that trigger multiple calcium-dependent intracellular responses that play key roles in learning, memory, and cognition. Excessive

NMDAR activation contributes to neuronal cell death in stroke, traumatic brain injury and various neurodegenerative diseases (Kamat et al., 2016; Koutsilieri and Riederer,

2007; Pivovarova and Andrews, 2010), whereas too little NMDAR activity impairs CNS function and, in particular, may cause symptoms seen in schizophrenia and autism

(Coyle, 2006; Kantrowitz and Javitt, 2010; Lisman et al., 2008). Thus, the recent development of agents that augment NMDAR activity (positive allosteric modulators, or

PAMs) offers an alternative approach for treating neuropsychiatric disorders such as schizophrenia that are not fully managed by currently available therapies. Of the genetic defects associated with schizophrenia, some would be expected to cause global NMDAR hypofunction – for example a defect in D- racemase (Luykx et al., 2015;

Schizophrenia Working Group of the Psychiatric Genomics, 2014), whereas other defects would be expected to affect subpopulations of NMDARs such as defects in genes which code for individual NMDAR subunits (Greenwood et al., 2012; Sun et al., 2010). Thus, global and subtype-specific NMDAR PAMs may each have patient-specific indications.

3 NMDAR complexes are composed of subunits from seven genes - GluN1,

GluN2A-GluN2D, and GluN3A-GluN3B (Ishii et al., 1993; Mishina et al., 1993; Monyer et al., 1994). These subunits assemble into hetero-tetrameric complexes in various combinations resulting in functionally-distinct NMDARs. Many NMDARs are thought to be composed of two GluN1 subunits and two GluN2 subunits. The different alternatively spliced GluN1 isoforms have largely similar pharmacological and physiological properties whereas the GluN2 subunits confer distinct physiological, biochemical, and pharmacological properties to the NMDAR complex (Buller et al., 1994; Hollmann et al.,

1993; Ikeda et al., 1992; Monyer et al., 1994; Sugihara et al., 1992; Vicini et al., 1998).

These properties, combined with their varied developmental profiles and anatomical distributions (Watanabe et al., 1992, 1993), imply that GluN2 subtype-selective agents would have distinct physiological and therapeutic properties.

Previously we have reported multiple aromatic ring structures substituted with a carboxylic acid group that display NMDAR PAM and/or NAM activity with varied patterns of subunit selectivity (Costa et al., 2012; Costa et al., 2010; Irvine et al., 2012;

Irvine et al., 2015). These agents are allosteric modulators interacting at the ligand binding domain (LBD) but they do not compete with either glutamate or glycine binding, nor do they bind at the N-terminal regulatory domain or within the ion channel (Costa et al., 2010). In contrast to agents that potentiate NMDARs containing specific GluN2 subunits, e.g. pregnenalone sulphate (PS) (Horak et al., 2006), UBP710 -

GluN2A/GluN2B; UBP512 - GluN2A (Costa et al., 2010); GNE-8324 - GluN2A

(Hackos et al., 2016); CIQ – GluN2C/GluN2D (Mullasseril et al., 2010); PYD-106 –

GluN2C (Khatri et al., 2014), the phenanthroic acid derivative UBP646 (Costa et al.,

4 2010) and the cholesterol derivative SGE-201 (Paul et al., 2013) potentiate all four

GluN1/GluN2 subtypes. Thus, in cases where it would be useful to augment global

NMDAR function, agents with these properties may be beneficial.

In this study, we characterize the functional properties of two naphthoic acid derivatives related to UBP646 which display improved effectiveness in enhancing currents at each of the four GluN1/GluN2 NMDARs, UBP684 (6-(4-methylpent-1-yl)-2- naphthoic acid) and UBP753 ((RS)-6-(5-methylhexan-2-yl)-2-naphthoic acid). We also identify mechanisms by which these agents can enhance NMDAR currents.

2. Methods

2.1 Compounds

UBP684, UBP753 and UBP792 ((E)-3-hydroxy-7-(2-nitrostyryl)-2-naphthoic acid) were synthesized and their structures were confirmed by 1H- and 13C-nuclear magnetic resonance (NMR) as well as mass spectroscopy. All compounds had elemental analyses where the determined percentage of C, H and N were less than 0.4 % different from theoretical values. Details of synthesis and purification will be reported elsewhere. Stock solutions were prepared in dimethyl sulfoxide at a concentration of 50 mM. The working solution was prepared in recording buffer just before the experiment. Other chemicals were obtained from Sigma unless stated otherwise.

2.2 GluN1 Subunit Expression in Xenopus oocytes

cDNA encoding the NMDAR GluN1-1a subunit was a generous gift of Dr.

Shigetada Nakanishi (Kyoto, Japan). cDNA encoding the GluNA, GluN2C and GluN2D subunits were kindly provided by Dr. Peter Seeburg (Heidelburg, Germany) and the

5 GluN2B [5’UTR] cDNA was the generous gift of Drs. Dolan Pritchett and David Lynch

(Philadelphia, USA). GluN1 and GluN2A constructs with cysteine substitution at N499C and Q686C in GluN1 (hereafter GluN1C) and at K487C and N687C in GluN2A (hereafter

GluN2AC) were kindly provided by Dr. Gabriela Popescu (University of Buffalo, USA).

Plasmids were linearized with Not I (GluN1a, GluN1C, GluN2AC), EcoR I (GluN2A,

GluN2C and GluN2D) or Sal I (GluN2B) and transcribed in vitro with T3 (GluN2A,

GluN2C), SP6 (GluN2B) or T7 (GluN1a, GluN2D) RNA polymerase using mMessage mMachine transcription kits (Ambion, Austin, TX, USA).

Oocytes were removed and isolated from mature female Xenopus laevis (Xenopus

One, Ann Arbor, MI, USA) as previously described (Buller et al., 1994). Procedures for animal handling were approved by the University of Nebraska Medical Center’s Animal

Care and Use Committee in compliance with the National Institutes of Health guidelines.

NMDAR subunit RNAs were dissolved in sterile distilled H2O. GluN1a and GluN2

RNAs were mixed in a molar ratio of 1:3. 50 nl of the final RNA mixture was microinjected (15-30 ng total) into the cytoplasm of oocyte. Oocytes were incubated in

ND-96 solution at 17°C prior to electrophysiological assay (1-5 days).

2.3 Two electrode voltage clamp electrophysiology

Electrophysiological responses were measured using a standard two-microelectrode voltage clamp (Warner Instruments, Hamden, Connecticut, model OC-725B) designed to provide fast clamp of large cells. The recording buffer contained (mM) 116 NaCl, 2 KCl,

0.3 BaCl2, 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.005

EGTA (or 0.01 diethylenetriaminepentaacetic acid, DTPA), and pH was adjusted to 7.4.

Response magnitude was determined by the steady-state plateau response elicited by bath

6 application of 10 µM L-glutamate plus 10 µM glycine at a holding potential of –60 mV unless stated otherwise. Response amplitudes for the four heteromeric complexes were generally between 0.2 to 1.5 µA. Compounds were bath applied in recording buffer

(Automate Scientific 8- or 16-channel perfusion system) and the responses were digitized for quantification (Digidata 1440A and pClamp-10, Molecular Devices). Dose-response relationships were fit to a single-site (GraphPad Prism, ISI Software, San Diego, CA,

USA), using a nonlinear regression to calculate IC50 or EC50 and % maximal response.

2.4 Hippocampal neuron whole-cell patch clamp recordings.

Whole-cell electrophysiology was conducted as previously described (Chopra et al.,

2015). Briefly, mice (~30-35 day old) were anesthetized by and decapitated in accordance with the approved protocols of Creighton University IACUC. The brain was rapidly removed and placed in ice-cold artificial cerebrospinal fluid (ACSF) of the following composition (in mM): 130 NaCl, 24 NaHCO3, 3.5 KCl, 1.25 NaH2PO4, 0.5

CaCl2, 3 MgCl2 and 10 glucose saturated with 95% O2/5% CO2. 300 µm thick sagittal sections were prepared using vibrating microtome (Leica VT1200, Buffalo Grove, IL,

USA). Whole-cell patch recordings were obtained from CA1 pyramidal neurons in voltage-clamp configuration at +40 mV with an Axopatch 200B (Molecular Devices,

Sunnyvale, CA, USA). Glass pipette with a resistance of 5–8 mOhm were filled with an internal solution consisting of (in mM) 110 cesium gluconate, 30 CsCl, 5 HEPES, 4 NaCl,

0.5 CaCl2, 2 MgCl2, 5 BAPTA,2 Na2ATP, and 0.3 Na2GTP (pH 7.35). The recording

ACSF contained (in mM) 1.5 CaCl2and 1.5 MgCl2. NMDAR responses were recorded in the presence of 0.5 µM tetrodotoxin, 100 µM and 10 µM NBQX. Signal was filtered at 2 kHz and digitized at 5 kHz using an Axon Digidata 1440A analog-to-digital

7 board (Molecular Devices, CA). NMDAR responses were obtained by briefly applying agonists (100 µM NMDA + 100 µM glycine) dissolved in the extracellular buffer using a

Picospritzer II. The application duration ranged from 30-50 ms. UBP684 (60 µM) was applied to the bath solution and changes in agonist responses were noted.

2.5 HEK cell patch-clamp recordings

Cell transfection and electrophysiology were performed as described previously (Bresink et al., 1996). Briefly, HEK 293 cells were transfected with GluN1a and GluN2A in the presence of 5 µg of eGFP (enhanced Green Fluorescent Protein) DNA in order to aid visualization of the transfected cells. Electrophysiological experiments were performed at room temperature. External bath solution contained (in mM): 145 NaCl, 2 KCl, 10

HEPES, 10 Glucose, 0.5 CaCl2, 0.01 EDTA, 0.05 Glycine. Internal pipette solution contained (in mM): 110 Cs Gluconate, 5 HEPES, 0.5 CaCl2, 2 Mg ATP, 0.3 mM Na

GTP, 30 CsCl2, 8 NaCl, 5 BAPTA; pH 7.35. Cells were visualized using a 20 x objective and phase contrast optics on an inverted microscope (Nikon, Japan). Epifluorescence of the cells expressing eGFP was excited using standard fluorescein filters and most of the cells expressing eGFP gave strong glutamate-mediated responses. Rapid application of glutamate and glycine was achieved using a 2-barreled theta glass pipette driven by a piezoelectric translator (Burleigh). NMDAR-mediated currents were recorded in either whole-cell or outside-out patch-clamp mode (Multiclamp 700A, Axon), digitized and stored on a PC for off-line analysis (Signal software, Cambridge Electronic Design

Limited).

2.6 Data analysis

8 The association time constant (KON) of UBP753, was calculated by determining the association time (τONSET) of L-glutamate/glycine-evoked current responses at different concentrations of UBP753 that were fit with a single exponential. Linear regression analysis of a plot of 1/ τONSET versus UBP753 concentration gave the KON (slope) and

KOFF (y-intercept) values which were used to calculate the KD (KD=KOFF/KON).

2.7 Statistical analyses

All values are expressed as mean ± SEM. Paired and unpaired t-test was used for comparing two numbers and comparisons with more than 2 groups were evaluated by one-way ANOVA followed by a Tukey’s or Bonferroni’s multiple comparison test.

Comparisons were considered as statistically significant if p < 0.05.

9 3. Results

3.1 The effects of agonists on PAM activity and of PAMs on agonist activity The ability of a PAM to potentiate NMDARs can depend upon the effect of agonist concentrations on PAM activity. And in a reciprocal manner, PAM binding can alter agonist activity. Thus, we determined the effect of different agonist concentrations on

PAM activity. UBP684 dose-response relationships were determined for the potentiation of GluN2A-D NMDAR responses evoked by 10 µM L-glutamate /10 µM glycine or by

300 µM L-glutamate / 300 µM glycine (Fig. 1, Table 1). UBP684 potentiated responses to low agonist concentrations at each of the NMDAR types with similar EC50s of approximately 30 µM and a maximal potentiation of 69 to 117 % (Table 1). In the presence of high agonist concentrations (Fig. 1C), UBP684 retained its ability to potentiate NMDAR responses. The degree of maximal potentiation was not significantly changed at GluN1a/GluN2A, GluN1a/GluN2C, and GluN1a/GluN2D receptor subtypes and was decreased by 40% at GluN1a/GluN2B receptors. High agonist concentrations enhanced UBP684 potency at receptors containing GluN2A and GluN2B subunits as reflected by a 63% and 28% reduction in EC50, respectively (Table 1), while UBP684 potency at the GluN2D subtype was lowered (93% increase in the EC50).

To determine if UBP684 has PAM activity at native NMDARs, we briefly applied

100 µM NMDA plus 100 µM glycine to CA1 pyramidal cells in hippocampal slices from

1 month-old mice. We found that bath application of UBP684 significantly increased the amplitude of agonist-induced currents (99.0 ± 34.7 % potentiation, p = 0.046, t-test) relative to the initial agonist response (Figure 1D). The potentiation was fully reversible with no detectible potentiation after UBP684 washout (-3.6 ± 8.3 % potentiation. The potentiated response was significantly different from the washout condition (p = 0.021).

10 To further define the effect of the PAMs on agonist responses, we determined the effect of 50 µM UBP684 on the dose-response relationship for L-glutamate and for glycine (Fig. 2, Table 2). Depending upon the subunits studied, PAM activity was associated with small shifts in agonist potencies as well as an increase in the maximal response to both agonists. UBP684 increased L-glutamate potency (32% reduction in L- glutamate EC50), but not glycine potency at GluN2A-containing receptors. In contrast, at

GluN2B-containing receptors, UBP684 increased glycine potency (30% reduction in glycine EC50), but not L-glutamate potency (Fig. 2 and Table 2). Since UBP684 increases glycine potency, then it is expected that UBP684 would increase GluN1/GluN2B responses more at low glycine concentrations than at high glycine concentrations. This is consistent with the partial reduction we observed in the maximum potentiation of

GluN1a/GluN2B responses by UBP684 when high agonist concentrations were used

(Table 1). In contrast, at GluN2C- and GluN2D-containing receptors, UBP684 reduced

L-glutamate potency (58% and 59% increase in EC50, respectively) and did not significantly change glycine potency (Table 2). Overall, and consistent with the low / high agonist concentration experiments, UBP684 increases the maximal effect of both agonists at all NMDARs at saturating agonist concentrations and additionally has minor, subtype-specific effects on agonist potencies.

UBP753 has an apparent potency that is similar to that of UBP684 and effectively potentiates agonist-induced responses on all four GluN1/GluN2 receptors (Table 1; Fig.

3). However, the limited solubility of UBP753 at concentrations above 100 µM made it difficult to establish saturating conditions to accurately define the EC50. To independently estimate potency, UBP753 on-rate and off-rates were determined at different

11 concentrations (Fig. 3B). The resulting rate constants indicated a Kd of 73 µM for

UBP753 that was 2-fold higher than the EC50 for UBP753 estimated by concentration- response analysis. Of relevance to other experiments described below (section 3.2), the slow, dose-dependent on-rates, and dose-independent off-rates for UBP753 indicate that

UBP753 binding/unbinding is significantly slower than the solution turnover time.

The ability of UBP684 to reduce L-glutamate potency at GluN1/GluN2D was unexpected for a PAM, although consistent with the greater PAM activity seen with high agonist concentrations (Table 1). Thus, we also evaluated UBP753, for its effect on agonist activity at GluN1a/GluN2D receptors. Like UBP684, UBP753 decreased L- glutamate potency (Fig. 3, Table 2) and had no effect on glycine potency. UBP753 increase